[evlatests] VLA archive oddities

[Walter Brisken]

I have modified idcaf and obs2script to temporarily work around the 
filename-induced project code problems.  Each source will come with the 
ObsCode as an intent that will be used by idcaf to label the data.  Thus 
projects that are passed through obs2script (all observing, most tests) 
will get this intent.  If the intent is not found the old demangling of 
the name will be attempted.

I realize this goes against the goal of not providing the obscode in 
multiple places (allowing for confusion), but in these cases where a large 
fraction of projects are getting their obscodes archived incorrectly I 
think this is safest mechanism.  The changest to each bit of code are very 
minor.  I will have these bits of code installed when convenient sometime 
today.

-Walter

On Fri, 29 Jun 2007, Ken Sowinski wrote:

> On Fri, 29 Jun 2007, Walter Brisken wrote:
>> 
>> The file run by the executor was: 000B124 .  IDCAF demangled this to
>> AB124.  The leading zero must have been added by the archive lister.
>> 
>> There is some file name limit somewhere that requires that the portion of
>> the filename preceding the . is 7 chars and the first three are reserved
>> for LST DAY mod 1000.  To make room the "A" is dropped and if a fourth
>> digit exists it will be dropped too.  Archive name problems will be
>> resolved soon.
>
> 1.  This is a consequence of choosing to use the filename to generate
> the program code rather than the program code on the ./ID card and
> not telling the operators how to create filenames.
>
> 2.  You must allow for four digit program code number parts.
> AB1234 should not be confused with AB123 or AB234.
>
> 3.  We need a file naming convention very soon that meets all
> our requirements and is understandable to the operators.
>